Improved Glycemic Outcomes With Diabetes Technology Use Independent of Socioeconomic Status in Youth With Type 1 Diabetes

Socioeconomic status (SES), ethnicity, and technology use are associated with glycemic outcomes in type 1 diabetes (T1D) (1–4). Furthermore, socioeconomic disadvantage may impact diabetes technology uptake and use (5,6), even when fully subsidized funding models exist (7,8). In Australia, continuous glucose monitoring (CGM) is fully subsidized for youth with T1D, but pump therapy is not. Therefore, the modifiable barrier of technology funding is critical for development of equitable models of care, given that models providing subsidized access to technology have been observed to partially mitigate SES disparities in technology uptake (9). The aim of this analysis was to evaluate whether SES modifies the association between technology use and improved glycemic control, as measured by HbA_1c, given that limited studies address this important question.

Cross-sectional analysis of HbA_1c was undertaken in youth with T1D receiving care at Australian pediatric diabetes centers contributing data to the Australasian Diabetes Data Network (ADDN) (10). Consent for data sharing with ADDN is obtained by an individual’s diabetes center. Approval for this analysis was gained through the Monash University human research ethics committee (project ID 38039).

Youth <18 years of age contributing data to ADDN between 1 July 2020 to 1 July 2022 were included. CGM and/or pump use records closest to 1 July 2022 were identified. A 2-year period was allowed for device use specification prior to 1 July 2022, given the impact of the coronavirus 2019 pandemic on data collection. Inclusion required an HbA_1c measurement within 6 months of device record and a recorded percent use of CGM over a 2-week period. Participants with CGM wear recorded as zero despite documentation of wearing a CGM device were reclassified as not using CGM.

Participants were excluded if no HbA_1c value within 6 months of the device use date was recorded or T1D diagnosis was within 6 months of the HbA_1c.

The Socio-Economic Indexes for Areas Index of Relative Socio-Economic Disadvantage (IRSD) uses Australian census data to rank geographical areas based on income, education level, employment type and status, spoken English ability, family structure, disability needs, housing, internet access, and vehicle factors (11). A lower ranking indicates greater socioeconomic disadvantage. The IRSD ranking of a participant’s most recent primary residential postcode was converted into quintiles. The most recent IRSD available at the time of analysis was based on 2016 Australian census data.

The Remoteness Areas Structure within the Australian Statistical Geography Standard classifies areas by remoteness as Australian major cities, inner regional, outer regional, remote, or very remote (12). Study participants were classified based on primary residential postcode as residing in an urban area (postcodes within an Australian major city) or remote area (all other postcodes).

Participant characteristics were summarized using mean (SD) and median (interquartile range [IQR]) for continuous measures and count (percentage) for categorical measures. Mean (SD) and median (IQR) HbA_1c are presented by IRSD quintile and by the four diabetes management regimens: insulin injection only, pump use only, CGM use only, and concurrent pump and CGM use.

To determine the independent associations of IRSD quintile and diabetes management regimen with HbA_1c, a linear regression model was conducted that adjusted for age at T1D diagnosis, T1D duration, sex, and remoteness classification. To determine whether associations between diabetes management regimen and HbA_1c differed by IRSD quintile, the model was expanded to include an interaction term (diabetes management regimen by IRSD quintile), and an F test comparing the base model to the expanded model was conducted.

Data from 2,822 youth were included, representing 49.7% (n = 5,683) of the ADDN data set as of 1 July 2022, excluding those diagnosed with T1D within 6 months (n = 230). Further individuals were excluded due to no clinical visit within 2 years (n = 798), missing percent CGM wear (n = 773), no HbA_1c within 2 years (n = 417), no HbA_1c within 6 months of device use (n = 549), HbA_1c within 6 months of T1D diagnosis (n = 180), and missing postcode data (n = 144). Of the final study cohort, 50.4% were female, mean (SD) age was 12.4 (3.6) years, mean diabetes duration was 5.1 (3.6) years, and mean HbA_1c was 8.1% (1.6%) (65 [17.5] mmol/mol) (Table 1). Study cohort characteristics were representative of the total ADDN population (Supplementary Table 1).

Of the study cohort, 699 (24.8%) were on insulin injections only, 449 (15.9%) were using a pump only, 649 (23%) were using CGM only (with insulin injections), and 1,025 (36.3%) were using a pump and CGM concurrently (manual mode or automated insulin delivery [AID]). Unadjusted mean and median HbA_1c by IRSD quintile and diabetes management regimen are presented in Table 2. Diabetes management regimen by IRSD quintile is shown in Supplementary Table 2.

After adjusting for age at diagnosis, T1D duration, sex, and remoteness classification, IRSD quintile (P < 0.001) and technology use (P < 0.001) were both independently associated with mean HbA_1c (Table 3). Mean HbA_1c was 0.51% lower (95% CI 0.69, 0.33) in the least disadvantaged IRSD quintile 5 compared with the most disadvantaged quintile 1. Compared with insulin injections only, pump use only was associated with a 0.45% lower mean HbA_1c (95% CI 0.66, 0.25), CGM use only (with insulin injections) with a 0.70% lower HbA_1c (95% CI 0.87, 0.53), and concurrent pump and CGM use with a 1.24% lower HbA_1c (95% CI 1.41, 1.08). IRSD quintile by diabetes technology use was added as an interaction term to the model, and no improvement in model fit was seen (F[12, 2,792] = 0.825, P = 0.624), thus providing no evidence that the association between technology use and HbA_1c differs by IRSD quintile. Figure 1 demonstrates this finding, presenting estimated marginal mean (EMM) HbA_1c percent (95% CI) across each IRSD quintile. Lower EMM HbA_1c was observed in those using any technology and especially for those concurrently using pump and CGM. This pattern was observed consistently across each IRSD quintile from the most disadvantaged quintile 1 to the least disadvantaged quintile 5.

In this cross-sectional analysis of youth with T1D, a differential association between technology use and mean HbA_1c by IRSD quintile was not observed. While causality cannot be inferred, the primary study finding is that regardless of relative socioeconomic disadvantage, all groups demonstrated better glycemic control with technology use. Barriers to technology access for those with greater socioeconomic disadvantage should be identified and addressed for the benefits of technology to be realized equitably.

Australia’s diabetes technology funding models are important to consider relative to the findings. In 2017, national subsidy enabled broad access to CGM for youth with T1D (13). Importantly, lower use of CGM in those with the greatest socioeconomic disadvantage persists in Australia (8). In contrast, pumps are primarily accessed through private health insurance funds or self-funding. Philanthropy provides limited additional access to those with socioeconomic disadvantage (14), but barriers to pump therapy access persist. Global analyses have outlined similar inequities in funding models for diabetes technology (15,16). Results from our analysis suggest that improving technology access may improve glycemic outcomes for youth with T1D, regardless of socioeconomic background.

Lower mean HbA_1c was observed with both pump and CGM use independently, with this observation being most pronounced for those using both pump and CGM. This relationship has been demonstrated in other studies; however, our novel finding is that improvement in HbA_1c with technology use is consistent across socioeconomic backgrounds. Supporting this finding, real-time CGM and pump use were reported to partially mediate the discrepancy observed in HbA_1c across socioeconomic groups (17). Addala et al. (18) showed an improved HbA_1c trajectory over 12 months with CGM start within 1 month of T1D diagnosis, irrespective of ethnicity or health insurance. Given that many nonmodifiable factors contribute to an individual’s SES, which can impact glycemic outcomes (4,19), our study suggests that supporting technology access for all is an actionable health system goal.

A key strength of this study is the inclusion of a large number of youth with T1D from a national diabetes database, representing approximately one in five Australian youth with T1D <20 years of age (20), despite the impact of the coronavirus 2019 pandemic on data collection and outpatient clinic attendance. Furthermore, a diverse range of individuals from varying socioeconomic backgrounds were included, as represented by numbers across each IRSD quintile. Limitations include the assumption that area-based socioeconomic deprivation measures based on postcode and IRSD quintile reflect an individual’s situation. Neither the proportion of individuals using AID with a pump and CGM concurrently nor an analysis by individual ADDN center could be explored. Potential variation across centers might explain the finding of a relatively high number of pump users not using CGM.

Our findings strengthen the call for equitable access models to diabetes technology in the Australian and global context. The glycemic benefit of diabetes technologies observed was similar across all socioeconomic quintiles and was greatest with concurrent use of pump and CGM. With rapidly accumulating evidence from trial and real-world settings that technology including AID improves glycemic control for youth with T1D, it is paramount that efforts are made to prioritize equitable access to diabetes technology for youth with T1D of all backgrounds.

Acknowledgments. This research was conducted using the ADDN registry. The authors are grateful to JDRF Australia, the Australian Research Council, and the children and young people with diabetes and their families who provided the data.

Funding. This research was supported by the Channel 7 Telethon Trust and JDRF Australia grant 5-SRA-2021-1088-M-X.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. K.E.L., C.E.T., M.B.A., G.J.S., A.H., E.Z., K.L.E., H.C., S.Z., A.J.J., J.H., M.I.d.B., T.W.J., and E.A.D. contributed to the development of this research, results interpretation, and editing of the manuscript and approved the final version of the manuscript. The ADDN Study Group contributed data to the ADDN registry. E.A.D. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.